thibaud commited on
Commit
61c1d8e
1 Parent(s): 3025765

Upload 11 files

Browse files
control_v11p_sd21_ade20k.yaml ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ model:
2
+ target: cldm.cldm.ControlLDM
3
+ params:
4
+ linear_start: 0.00085
5
+ linear_end: 0.0120
6
+ num_timesteps_cond: 1
7
+ log_every_t: 200
8
+ timesteps: 1000
9
+ first_stage_key: "jpg"
10
+ cond_stage_key: "txt"
11
+ control_key: "hint"
12
+ image_size: 64
13
+ channels: 4
14
+ cond_stage_trainable: false
15
+ conditioning_key: crossattn
16
+ monitor: val/loss_simple_ema
17
+ scale_factor: 0.18215
18
+ use_ema: False
19
+ only_mid_control: False
20
+
21
+ control_stage_config:
22
+ target: cldm.cldm.ControlNet
23
+ params:
24
+ image_size: 32 # unused
25
+ in_channels: 4
26
+ hint_channels: 3
27
+ model_channels: 320
28
+ attention_resolutions: [ 4, 2, 1 ]
29
+ num_res_blocks: 2
30
+ channel_mult: [ 1, 2, 4, 4 ]
31
+ num_heads: 8
32
+ use_spatial_transformer: True
33
+ transformer_depth: 1
34
+ context_dim: 768
35
+ use_checkpoint: True
36
+ legacy: False
37
+
38
+ unet_config:
39
+ target: cldm.cldm.ControlledUnetModel
40
+ params:
41
+ image_size: 32 # unused
42
+ in_channels: 4
43
+ out_channels: 4
44
+ model_channels: 320
45
+ attention_resolutions: [ 4, 2, 1 ]
46
+ num_res_blocks: 2
47
+ channel_mult: [ 1, 2, 4, 4 ]
48
+ num_heads: 8
49
+ use_spatial_transformer: True
50
+ transformer_depth: 1
51
+ context_dim: 768
52
+ use_checkpoint: True
53
+ legacy: False
54
+
55
+ first_stage_config:
56
+ target: ldm.models.autoencoder.AutoencoderKL
57
+ params:
58
+ embed_dim: 4
59
+ monitor: val/rec_loss
60
+ ddconfig:
61
+ double_z: true
62
+ z_channels: 4
63
+ resolution: 256
64
+ in_channels: 3
65
+ out_ch: 3
66
+ ch: 128
67
+ ch_mult:
68
+ - 1
69
+ - 2
70
+ - 4
71
+ - 4
72
+ num_res_blocks: 2
73
+ attn_resolutions: []
74
+ dropout: 0.0
75
+ lossconfig:
76
+ target: torch.nn.Identity
77
+
78
+ cond_stage_config:
79
+ target: ldm.modules.encoders.modules.FrozenCLIPEmbedder
control_v11p_sd21_canny.yaml ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ model:
2
+ target: cldm.cldm.ControlLDM
3
+ params:
4
+ linear_start: 0.00085
5
+ linear_end: 0.0120
6
+ num_timesteps_cond: 1
7
+ log_every_t: 200
8
+ timesteps: 1000
9
+ first_stage_key: "jpg"
10
+ cond_stage_key: "txt"
11
+ control_key: "hint"
12
+ image_size: 64
13
+ channels: 4
14
+ cond_stage_trainable: false
15
+ conditioning_key: crossattn
16
+ monitor: val/loss_simple_ema
17
+ scale_factor: 0.18215
18
+ use_ema: False
19
+ only_mid_control: False
20
+
21
+ control_stage_config:
22
+ target: cldm.cldm.ControlNet
23
+ params:
24
+ image_size: 32 # unused
25
+ in_channels: 4
26
+ hint_channels: 3
27
+ model_channels: 320
28
+ attention_resolutions: [ 4, 2, 1 ]
29
+ num_res_blocks: 2
30
+ channel_mult: [ 1, 2, 4, 4 ]
31
+ num_heads: 8
32
+ use_spatial_transformer: True
33
+ transformer_depth: 1
34
+ context_dim: 768
35
+ use_checkpoint: True
36
+ legacy: False
37
+
38
+ unet_config:
39
+ target: cldm.cldm.ControlledUnetModel
40
+ params:
41
+ image_size: 32 # unused
42
+ in_channels: 4
43
+ out_channels: 4
44
+ model_channels: 320
45
+ attention_resolutions: [ 4, 2, 1 ]
46
+ num_res_blocks: 2
47
+ channel_mult: [ 1, 2, 4, 4 ]
48
+ num_heads: 8
49
+ use_spatial_transformer: True
50
+ transformer_depth: 1
51
+ context_dim: 768
52
+ use_checkpoint: True
53
+ legacy: False
54
+
55
+ first_stage_config:
56
+ target: ldm.models.autoencoder.AutoencoderKL
57
+ params:
58
+ embed_dim: 4
59
+ monitor: val/rec_loss
60
+ ddconfig:
61
+ double_z: true
62
+ z_channels: 4
63
+ resolution: 256
64
+ in_channels: 3
65
+ out_ch: 3
66
+ ch: 128
67
+ ch_mult:
68
+ - 1
69
+ - 2
70
+ - 4
71
+ - 4
72
+ num_res_blocks: 2
73
+ attn_resolutions: []
74
+ dropout: 0.0
75
+ lossconfig:
76
+ target: torch.nn.Identity
77
+
78
+ cond_stage_config:
79
+ target: ldm.modules.encoders.modules.FrozenCLIPEmbedder
control_v11p_sd21_color.yaml ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ model:
2
+ target: cldm.cldm.ControlLDM
3
+ params:
4
+ linear_start: 0.00085
5
+ linear_end: 0.0120
6
+ num_timesteps_cond: 1
7
+ log_every_t: 200
8
+ timesteps: 1000
9
+ first_stage_key: "jpg"
10
+ cond_stage_key: "txt"
11
+ control_key: "hint"
12
+ image_size: 64
13
+ channels: 4
14
+ cond_stage_trainable: false
15
+ conditioning_key: crossattn
16
+ monitor: val/loss_simple_ema
17
+ scale_factor: 0.18215
18
+ use_ema: False
19
+ only_mid_control: False
20
+
21
+ control_stage_config:
22
+ target: cldm.cldm.ControlNet
23
+ params:
24
+ image_size: 32 # unused
25
+ in_channels: 4
26
+ hint_channels: 3
27
+ model_channels: 320
28
+ attention_resolutions: [ 4, 2, 1 ]
29
+ num_res_blocks: 2
30
+ channel_mult: [ 1, 2, 4, 4 ]
31
+ num_heads: 8
32
+ use_spatial_transformer: True
33
+ transformer_depth: 1
34
+ context_dim: 768
35
+ use_checkpoint: True
36
+ legacy: False
37
+
38
+ unet_config:
39
+ target: cldm.cldm.ControlledUnetModel
40
+ params:
41
+ image_size: 32 # unused
42
+ in_channels: 4
43
+ out_channels: 4
44
+ model_channels: 320
45
+ attention_resolutions: [ 4, 2, 1 ]
46
+ num_res_blocks: 2
47
+ channel_mult: [ 1, 2, 4, 4 ]
48
+ num_heads: 8
49
+ use_spatial_transformer: True
50
+ transformer_depth: 1
51
+ context_dim: 768
52
+ use_checkpoint: True
53
+ legacy: False
54
+
55
+ first_stage_config:
56
+ target: ldm.models.autoencoder.AutoencoderKL
57
+ params:
58
+ embed_dim: 4
59
+ monitor: val/rec_loss
60
+ ddconfig:
61
+ double_z: true
62
+ z_channels: 4
63
+ resolution: 256
64
+ in_channels: 3
65
+ out_ch: 3
66
+ ch: 128
67
+ ch_mult:
68
+ - 1
69
+ - 2
70
+ - 4
71
+ - 4
72
+ num_res_blocks: 2
73
+ attn_resolutions: []
74
+ dropout: 0.0
75
+ lossconfig:
76
+ target: torch.nn.Identity
77
+
78
+ cond_stage_config:
79
+ target: ldm.modules.encoders.modules.FrozenCLIPEmbedder
control_v11p_sd21_depth.yaml ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ model:
2
+ target: cldm.cldm.ControlLDM
3
+ params:
4
+ linear_start: 0.00085
5
+ linear_end: 0.0120
6
+ num_timesteps_cond: 1
7
+ log_every_t: 200
8
+ timesteps: 1000
9
+ first_stage_key: "jpg"
10
+ cond_stage_key: "txt"
11
+ control_key: "hint"
12
+ image_size: 64
13
+ channels: 4
14
+ cond_stage_trainable: false
15
+ conditioning_key: crossattn
16
+ monitor: val/loss_simple_ema
17
+ scale_factor: 0.18215
18
+ use_ema: False
19
+ only_mid_control: False
20
+
21
+ control_stage_config:
22
+ target: cldm.cldm.ControlNet
23
+ params:
24
+ image_size: 32 # unused
25
+ in_channels: 4
26
+ hint_channels: 3
27
+ model_channels: 320
28
+ attention_resolutions: [ 4, 2, 1 ]
29
+ num_res_blocks: 2
30
+ channel_mult: [ 1, 2, 4, 4 ]
31
+ num_heads: 8
32
+ use_spatial_transformer: True
33
+ transformer_depth: 1
34
+ context_dim: 768
35
+ use_checkpoint: True
36
+ legacy: False
37
+
38
+ unet_config:
39
+ target: cldm.cldm.ControlledUnetModel
40
+ params:
41
+ image_size: 32 # unused
42
+ in_channels: 4
43
+ out_channels: 4
44
+ model_channels: 320
45
+ attention_resolutions: [ 4, 2, 1 ]
46
+ num_res_blocks: 2
47
+ channel_mult: [ 1, 2, 4, 4 ]
48
+ num_heads: 8
49
+ use_spatial_transformer: True
50
+ transformer_depth: 1
51
+ context_dim: 768
52
+ use_checkpoint: True
53
+ legacy: False
54
+
55
+ first_stage_config:
56
+ target: ldm.models.autoencoder.AutoencoderKL
57
+ params:
58
+ embed_dim: 4
59
+ monitor: val/rec_loss
60
+ ddconfig:
61
+ double_z: true
62
+ z_channels: 4
63
+ resolution: 256
64
+ in_channels: 3
65
+ out_ch: 3
66
+ ch: 128
67
+ ch_mult:
68
+ - 1
69
+ - 2
70
+ - 4
71
+ - 4
72
+ num_res_blocks: 2
73
+ attn_resolutions: []
74
+ dropout: 0.0
75
+ lossconfig:
76
+ target: torch.nn.Identity
77
+
78
+ cond_stage_config:
79
+ target: ldm.modules.encoders.modules.FrozenCLIPEmbedder
control_v11p_sd21_hed.yaml ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ model:
2
+ target: cldm.cldm.ControlLDM
3
+ params:
4
+ linear_start: 0.00085
5
+ linear_end: 0.0120
6
+ num_timesteps_cond: 1
7
+ log_every_t: 200
8
+ timesteps: 1000
9
+ first_stage_key: "jpg"
10
+ cond_stage_key: "txt"
11
+ control_key: "hint"
12
+ image_size: 64
13
+ channels: 4
14
+ cond_stage_trainable: false
15
+ conditioning_key: crossattn
16
+ monitor: val/loss_simple_ema
17
+ scale_factor: 0.18215
18
+ use_ema: False
19
+ only_mid_control: False
20
+
21
+ control_stage_config:
22
+ target: cldm.cldm.ControlNet
23
+ params:
24
+ image_size: 32 # unused
25
+ in_channels: 4
26
+ hint_channels: 3
27
+ model_channels: 320
28
+ attention_resolutions: [ 4, 2, 1 ]
29
+ num_res_blocks: 2
30
+ channel_mult: [ 1, 2, 4, 4 ]
31
+ num_heads: 8
32
+ use_spatial_transformer: True
33
+ transformer_depth: 1
34
+ context_dim: 768
35
+ use_checkpoint: True
36
+ legacy: False
37
+
38
+ unet_config:
39
+ target: cldm.cldm.ControlledUnetModel
40
+ params:
41
+ image_size: 32 # unused
42
+ in_channels: 4
43
+ out_channels: 4
44
+ model_channels: 320
45
+ attention_resolutions: [ 4, 2, 1 ]
46
+ num_res_blocks: 2
47
+ channel_mult: [ 1, 2, 4, 4 ]
48
+ num_heads: 8
49
+ use_spatial_transformer: True
50
+ transformer_depth: 1
51
+ context_dim: 768
52
+ use_checkpoint: True
53
+ legacy: False
54
+
55
+ first_stage_config:
56
+ target: ldm.models.autoencoder.AutoencoderKL
57
+ params:
58
+ embed_dim: 4
59
+ monitor: val/rec_loss
60
+ ddconfig:
61
+ double_z: true
62
+ z_channels: 4
63
+ resolution: 256
64
+ in_channels: 3
65
+ out_ch: 3
66
+ ch: 128
67
+ ch_mult:
68
+ - 1
69
+ - 2
70
+ - 4
71
+ - 4
72
+ num_res_blocks: 2
73
+ attn_resolutions: []
74
+ dropout: 0.0
75
+ lossconfig:
76
+ target: torch.nn.Identity
77
+
78
+ cond_stage_config:
79
+ target: ldm.modules.encoders.modules.FrozenCLIPEmbedder
control_v11p_sd21_lineart.yaml ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ model:
2
+ target: cldm.cldm.ControlLDM
3
+ params:
4
+ linear_start: 0.00085
5
+ linear_end: 0.0120
6
+ num_timesteps_cond: 1
7
+ log_every_t: 200
8
+ timesteps: 1000
9
+ first_stage_key: "jpg"
10
+ cond_stage_key: "txt"
11
+ control_key: "hint"
12
+ image_size: 64
13
+ channels: 4
14
+ cond_stage_trainable: false
15
+ conditioning_key: crossattn
16
+ monitor: val/loss_simple_ema
17
+ scale_factor: 0.18215
18
+ use_ema: False
19
+ only_mid_control: False
20
+
21
+ control_stage_config:
22
+ target: cldm.cldm.ControlNet
23
+ params:
24
+ image_size: 32 # unused
25
+ in_channels: 4
26
+ hint_channels: 3
27
+ model_channels: 320
28
+ attention_resolutions: [ 4, 2, 1 ]
29
+ num_res_blocks: 2
30
+ channel_mult: [ 1, 2, 4, 4 ]
31
+ num_heads: 8
32
+ use_spatial_transformer: True
33
+ transformer_depth: 1
34
+ context_dim: 768
35
+ use_checkpoint: True
36
+ legacy: False
37
+
38
+ unet_config:
39
+ target: cldm.cldm.ControlledUnetModel
40
+ params:
41
+ image_size: 32 # unused
42
+ in_channels: 4
43
+ out_channels: 4
44
+ model_channels: 320
45
+ attention_resolutions: [ 4, 2, 1 ]
46
+ num_res_blocks: 2
47
+ channel_mult: [ 1, 2, 4, 4 ]
48
+ num_heads: 8
49
+ use_spatial_transformer: True
50
+ transformer_depth: 1
51
+ context_dim: 768
52
+ use_checkpoint: True
53
+ legacy: False
54
+
55
+ first_stage_config:
56
+ target: ldm.models.autoencoder.AutoencoderKL
57
+ params:
58
+ embed_dim: 4
59
+ monitor: val/rec_loss
60
+ ddconfig:
61
+ double_z: true
62
+ z_channels: 4
63
+ resolution: 256
64
+ in_channels: 3
65
+ out_ch: 3
66
+ ch: 128
67
+ ch_mult:
68
+ - 1
69
+ - 2
70
+ - 4
71
+ - 4
72
+ num_res_blocks: 2
73
+ attn_resolutions: []
74
+ dropout: 0.0
75
+ lossconfig:
76
+ target: torch.nn.Identity
77
+
78
+ cond_stage_config:
79
+ target: ldm.modules.encoders.modules.FrozenCLIPEmbedder
control_v11p_sd21_normalbae.yaml ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ model:
2
+ target: cldm.cldm.ControlLDM
3
+ params:
4
+ linear_start: 0.00085
5
+ linear_end: 0.0120
6
+ num_timesteps_cond: 1
7
+ log_every_t: 200
8
+ timesteps: 1000
9
+ first_stage_key: "jpg"
10
+ cond_stage_key: "txt"
11
+ control_key: "hint"
12
+ image_size: 64
13
+ channels: 4
14
+ cond_stage_trainable: false
15
+ conditioning_key: crossattn
16
+ monitor: val/loss_simple_ema
17
+ scale_factor: 0.18215
18
+ use_ema: False
19
+ only_mid_control: False
20
+
21
+ control_stage_config:
22
+ target: cldm.cldm.ControlNet
23
+ params:
24
+ image_size: 32 # unused
25
+ in_channels: 4
26
+ hint_channels: 3
27
+ model_channels: 320
28
+ attention_resolutions: [ 4, 2, 1 ]
29
+ num_res_blocks: 2
30
+ channel_mult: [ 1, 2, 4, 4 ]
31
+ num_heads: 8
32
+ use_spatial_transformer: True
33
+ transformer_depth: 1
34
+ context_dim: 768
35
+ use_checkpoint: True
36
+ legacy: False
37
+
38
+ unet_config:
39
+ target: cldm.cldm.ControlledUnetModel
40
+ params:
41
+ image_size: 32 # unused
42
+ in_channels: 4
43
+ out_channels: 4
44
+ model_channels: 320
45
+ attention_resolutions: [ 4, 2, 1 ]
46
+ num_res_blocks: 2
47
+ channel_mult: [ 1, 2, 4, 4 ]
48
+ num_heads: 8
49
+ use_spatial_transformer: True
50
+ transformer_depth: 1
51
+ context_dim: 768
52
+ use_checkpoint: True
53
+ legacy: False
54
+
55
+ first_stage_config:
56
+ target: ldm.models.autoencoder.AutoencoderKL
57
+ params:
58
+ embed_dim: 4
59
+ monitor: val/rec_loss
60
+ ddconfig:
61
+ double_z: true
62
+ z_channels: 4
63
+ resolution: 256
64
+ in_channels: 3
65
+ out_ch: 3
66
+ ch: 128
67
+ ch_mult:
68
+ - 1
69
+ - 2
70
+ - 4
71
+ - 4
72
+ num_res_blocks: 2
73
+ attn_resolutions: []
74
+ dropout: 0.0
75
+ lossconfig:
76
+ target: torch.nn.Identity
77
+
78
+ cond_stage_config:
79
+ target: ldm.modules.encoders.modules.FrozenCLIPEmbedder
control_v11p_sd21_openpose.yaml ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ model:
2
+ target: cldm.cldm.ControlLDM
3
+ params:
4
+ linear_start: 0.00085
5
+ linear_end: 0.0120
6
+ num_timesteps_cond: 1
7
+ log_every_t: 200
8
+ timesteps: 1000
9
+ first_stage_key: "jpg"
10
+ cond_stage_key: "txt"
11
+ control_key: "hint"
12
+ image_size: 64
13
+ channels: 4
14
+ cond_stage_trainable: false
15
+ conditioning_key: crossattn
16
+ monitor: val/loss_simple_ema
17
+ scale_factor: 0.18215
18
+ use_ema: False
19
+ only_mid_control: False
20
+
21
+ control_stage_config:
22
+ target: cldm.cldm.ControlNet
23
+ params:
24
+ image_size: 32 # unused
25
+ in_channels: 4
26
+ hint_channels: 3
27
+ model_channels: 320
28
+ attention_resolutions: [ 4, 2, 1 ]
29
+ num_res_blocks: 2
30
+ channel_mult: [ 1, 2, 4, 4 ]
31
+ num_heads: 8
32
+ use_spatial_transformer: True
33
+ transformer_depth: 1
34
+ context_dim: 768
35
+ use_checkpoint: True
36
+ legacy: False
37
+
38
+ unet_config:
39
+ target: cldm.cldm.ControlledUnetModel
40
+ params:
41
+ image_size: 32 # unused
42
+ in_channels: 4
43
+ out_channels: 4
44
+ model_channels: 320
45
+ attention_resolutions: [ 4, 2, 1 ]
46
+ num_res_blocks: 2
47
+ channel_mult: [ 1, 2, 4, 4 ]
48
+ num_heads: 8
49
+ use_spatial_transformer: True
50
+ transformer_depth: 1
51
+ context_dim: 768
52
+ use_checkpoint: True
53
+ legacy: False
54
+
55
+ first_stage_config:
56
+ target: ldm.models.autoencoder.AutoencoderKL
57
+ params:
58
+ embed_dim: 4
59
+ monitor: val/rec_loss
60
+ ddconfig:
61
+ double_z: true
62
+ z_channels: 4
63
+ resolution: 256
64
+ in_channels: 3
65
+ out_ch: 3
66
+ ch: 128
67
+ ch_mult:
68
+ - 1
69
+ - 2
70
+ - 4
71
+ - 4
72
+ num_res_blocks: 2
73
+ attn_resolutions: []
74
+ dropout: 0.0
75
+ lossconfig:
76
+ target: torch.nn.Identity
77
+
78
+ cond_stage_config:
79
+ target: ldm.modules.encoders.modules.FrozenCLIPEmbedder
control_v11p_sd21_openposev2.yaml ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ model:
2
+ target: cldm.cldm.ControlLDM
3
+ params:
4
+ linear_start: 0.00085
5
+ linear_end: 0.0120
6
+ num_timesteps_cond: 1
7
+ log_every_t: 200
8
+ timesteps: 1000
9
+ first_stage_key: "jpg"
10
+ cond_stage_key: "txt"
11
+ control_key: "hint"
12
+ image_size: 64
13
+ channels: 4
14
+ cond_stage_trainable: false
15
+ conditioning_key: crossattn
16
+ monitor: val/loss_simple_ema
17
+ scale_factor: 0.18215
18
+ use_ema: False
19
+ only_mid_control: False
20
+
21
+ control_stage_config:
22
+ target: cldm.cldm.ControlNet
23
+ params:
24
+ image_size: 32 # unused
25
+ in_channels: 4
26
+ hint_channels: 3
27
+ model_channels: 320
28
+ attention_resolutions: [ 4, 2, 1 ]
29
+ num_res_blocks: 2
30
+ channel_mult: [ 1, 2, 4, 4 ]
31
+ num_heads: 8
32
+ use_spatial_transformer: True
33
+ transformer_depth: 1
34
+ context_dim: 768
35
+ use_checkpoint: True
36
+ legacy: False
37
+
38
+ unet_config:
39
+ target: cldm.cldm.ControlledUnetModel
40
+ params:
41
+ image_size: 32 # unused
42
+ in_channels: 4
43
+ out_channels: 4
44
+ model_channels: 320
45
+ attention_resolutions: [ 4, 2, 1 ]
46
+ num_res_blocks: 2
47
+ channel_mult: [ 1, 2, 4, 4 ]
48
+ num_heads: 8
49
+ use_spatial_transformer: True
50
+ transformer_depth: 1
51
+ context_dim: 768
52
+ use_checkpoint: True
53
+ legacy: False
54
+
55
+ first_stage_config:
56
+ target: ldm.models.autoencoder.AutoencoderKL
57
+ params:
58
+ embed_dim: 4
59
+ monitor: val/rec_loss
60
+ ddconfig:
61
+ double_z: true
62
+ z_channels: 4
63
+ resolution: 256
64
+ in_channels: 3
65
+ out_ch: 3
66
+ ch: 128
67
+ ch_mult:
68
+ - 1
69
+ - 2
70
+ - 4
71
+ - 4
72
+ num_res_blocks: 2
73
+ attn_resolutions: []
74
+ dropout: 0.0
75
+ lossconfig:
76
+ target: torch.nn.Identity
77
+
78
+ cond_stage_config:
79
+ target: ldm.modules.encoders.modules.FrozenCLIPEmbedder
control_v11p_sd21_scribble.yaml ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ model:
2
+ target: cldm.cldm.ControlLDM
3
+ params:
4
+ linear_start: 0.00085
5
+ linear_end: 0.0120
6
+ num_timesteps_cond: 1
7
+ log_every_t: 200
8
+ timesteps: 1000
9
+ first_stage_key: "jpg"
10
+ cond_stage_key: "txt"
11
+ control_key: "hint"
12
+ image_size: 64
13
+ channels: 4
14
+ cond_stage_trainable: false
15
+ conditioning_key: crossattn
16
+ monitor: val/loss_simple_ema
17
+ scale_factor: 0.18215
18
+ use_ema: False
19
+ only_mid_control: False
20
+
21
+ control_stage_config:
22
+ target: cldm.cldm.ControlNet
23
+ params:
24
+ image_size: 32 # unused
25
+ in_channels: 4
26
+ hint_channels: 3
27
+ model_channels: 320
28
+ attention_resolutions: [ 4, 2, 1 ]
29
+ num_res_blocks: 2
30
+ channel_mult: [ 1, 2, 4, 4 ]
31
+ num_heads: 8
32
+ use_spatial_transformer: True
33
+ transformer_depth: 1
34
+ context_dim: 768
35
+ use_checkpoint: True
36
+ legacy: False
37
+
38
+ unet_config:
39
+ target: cldm.cldm.ControlledUnetModel
40
+ params:
41
+ image_size: 32 # unused
42
+ in_channels: 4
43
+ out_channels: 4
44
+ model_channels: 320
45
+ attention_resolutions: [ 4, 2, 1 ]
46
+ num_res_blocks: 2
47
+ channel_mult: [ 1, 2, 4, 4 ]
48
+ num_heads: 8
49
+ use_spatial_transformer: True
50
+ transformer_depth: 1
51
+ context_dim: 768
52
+ use_checkpoint: True
53
+ legacy: False
54
+
55
+ first_stage_config:
56
+ target: ldm.models.autoencoder.AutoencoderKL
57
+ params:
58
+ embed_dim: 4
59
+ monitor: val/rec_loss
60
+ ddconfig:
61
+ double_z: true
62
+ z_channels: 4
63
+ resolution: 256
64
+ in_channels: 3
65
+ out_ch: 3
66
+ ch: 128
67
+ ch_mult:
68
+ - 1
69
+ - 2
70
+ - 4
71
+ - 4
72
+ num_res_blocks: 2
73
+ attn_resolutions: []
74
+ dropout: 0.0
75
+ lossconfig:
76
+ target: torch.nn.Identity
77
+
78
+ cond_stage_config:
79
+ target: ldm.modules.encoders.modules.FrozenCLIPEmbedder
control_v11p_sd21_zoedepth.yaml ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ model:
2
+ target: cldm.cldm.ControlLDM
3
+ params:
4
+ linear_start: 0.00085
5
+ linear_end: 0.0120
6
+ num_timesteps_cond: 1
7
+ log_every_t: 200
8
+ timesteps: 1000
9
+ first_stage_key: "jpg"
10
+ cond_stage_key: "txt"
11
+ control_key: "hint"
12
+ image_size: 64
13
+ channels: 4
14
+ cond_stage_trainable: false
15
+ conditioning_key: crossattn
16
+ monitor: val/loss_simple_ema
17
+ scale_factor: 0.18215
18
+ use_ema: False
19
+ only_mid_control: False
20
+
21
+ control_stage_config:
22
+ target: cldm.cldm.ControlNet
23
+ params:
24
+ image_size: 32 # unused
25
+ in_channels: 4
26
+ hint_channels: 3
27
+ model_channels: 320
28
+ attention_resolutions: [ 4, 2, 1 ]
29
+ num_res_blocks: 2
30
+ channel_mult: [ 1, 2, 4, 4 ]
31
+ num_heads: 8
32
+ use_spatial_transformer: True
33
+ transformer_depth: 1
34
+ context_dim: 768
35
+ use_checkpoint: True
36
+ legacy: False
37
+
38
+ unet_config:
39
+ target: cldm.cldm.ControlledUnetModel
40
+ params:
41
+ image_size: 32 # unused
42
+ in_channels: 4
43
+ out_channels: 4
44
+ model_channels: 320
45
+ attention_resolutions: [ 4, 2, 1 ]
46
+ num_res_blocks: 2
47
+ channel_mult: [ 1, 2, 4, 4 ]
48
+ num_heads: 8
49
+ use_spatial_transformer: True
50
+ transformer_depth: 1
51
+ context_dim: 768
52
+ use_checkpoint: True
53
+ legacy: False
54
+
55
+ first_stage_config:
56
+ target: ldm.models.autoencoder.AutoencoderKL
57
+ params:
58
+ embed_dim: 4
59
+ monitor: val/rec_loss
60
+ ddconfig:
61
+ double_z: true
62
+ z_channels: 4
63
+ resolution: 256
64
+ in_channels: 3
65
+ out_ch: 3
66
+ ch: 128
67
+ ch_mult:
68
+ - 1
69
+ - 2
70
+ - 4
71
+ - 4
72
+ num_res_blocks: 2
73
+ attn_resolutions: []
74
+ dropout: 0.0
75
+ lossconfig:
76
+ target: torch.nn.Identity
77
+
78
+ cond_stage_config:
79
+ target: ldm.modules.encoders.modules.FrozenCLIPEmbedder